PAC-Bayesian Theorems for Domain Adaptation with Specialization to Linear Classifiers

TL;DR

This paper advances PAC-Bayesian theory for domain adaptation by introducing a tighter bound based on a novel distribution pseudodistance, specializing to linear classifiers and extending to multisource scenarios, with promising experimental results.

Contribution

It proposes a new PAC-Bayesian domain adaptation bound, improves previous methods, and extends the framework to multisource adaptation with practical algorithms.

Findings

Tighter PAC-Bayesian bound for domain adaptation.

Effective linear classifier learning algorithm demonstrated.

Successful experiments on synthetic and sentiment analysis tasks.

Abstract

In this paper, we provide two main contributions in PAC-Bayesian theory for domain adaptation where the objective is to learn, from a source distribution, a well-performing majority vote on a different target distribution. On the one hand, we propose an improvement of the previous approach proposed by Germain et al. (2013), that relies on a novel distribution pseudodistance based on a disagreement averaging, allowing us to derive a new tighter PAC-Bayesian domain adaptation bound for the stochastic Gibbs classifier. We specialize it to linear classifiers, and design a learning algorithm which shows interesting results on a synthetic problem and on a popular sentiment annotation task. On the other hand, we generalize these results to multisource domain adaptation allowing us to take into account different source domains. This study opens the door to tackle domain adaptation tasks by making use of all the PAC-Bayesian tools.